Distributed Search

AutoTorch provide seamless experience on single machine and distributed training. Simply providing the ip-addresses for your remote machines, AutoTorch will automatically schedule the job for you.

Goals of this tutorial:

• Understand how to conduct large-scale distributed search using AutoTorch.

• Use AWS SageMaker to schedule large scale experiments (TODO).

• Distributed Search on Dedicated Machines
  • Quick Example
  • System Implementation Logic
  • Distributed Training Setup on AWS EC2